1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* File containing the {@see Mailcode_Parser_Statement_Tokenizer} class. |
4
|
|
|
* |
5
|
|
|
* @package Mailcode |
6
|
|
|
* @subpackage Parser |
7
|
|
|
* @see Mailcode_Parser_Statement_Tokenizer |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
declare(strict_types=1); |
11
|
|
|
|
12
|
|
|
namespace Mailcode; |
13
|
|
|
|
14
|
|
|
/** |
15
|
|
|
* Mailcode statement tokenizer: parses a mailcode statement |
16
|
|
|
* into its logical parts. |
17
|
|
|
* |
18
|
|
|
* @package Mailcode |
19
|
|
|
* @subpackage Parser |
20
|
|
|
* @author Sebastian Mordziol <[email protected]> |
21
|
|
|
*/ |
22
|
|
|
class Mailcode_Parser_Statement_Tokenizer |
23
|
|
|
{ |
24
|
|
|
const ERROR_TOKENIZE_METHOD_MISSING = 49801; |
25
|
|
|
const ERROR_INVALID_TOKEN_CREATED = 49802; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* @var string[] |
29
|
|
|
*/ |
30
|
|
|
protected $tokenCategories = array( |
31
|
|
|
'Variables', |
32
|
|
|
'NormalizeQuotes', |
33
|
|
|
'EscapedQuotes', |
34
|
|
|
'StringLiterals', |
35
|
|
|
'Keywords', |
36
|
|
|
'Numbers', |
37
|
|
|
'Operands', |
38
|
|
|
'ExtractTokens' |
39
|
|
|
); |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @var Mailcode_Parser_Statement |
43
|
|
|
*/ |
44
|
|
|
protected $statement; |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* @var string |
48
|
|
|
*/ |
49
|
|
|
protected $tokenized; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* @var Mailcode_Parser_Statement_Tokenizer_Token[] |
53
|
|
|
*/ |
54
|
|
|
protected $tokensOrdered = array(); |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* @var string[] |
58
|
|
|
*/ |
59
|
|
|
protected static $ids = array(); |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @var callable[] |
63
|
|
|
*/ |
64
|
|
|
protected $changeHandlers = array(); |
65
|
|
|
|
66
|
|
|
public function __construct(Mailcode_Parser_Statement $statement) |
67
|
|
|
{ |
68
|
|
|
$this->statement = $statement; |
69
|
|
|
|
70
|
|
|
$this->tokenize($statement->getStatementString()); |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
public function getSourceCommand() : ?Mailcode_Commands_Command |
74
|
|
|
{ |
75
|
|
|
return $this->statement->getSourceCommand(); |
76
|
|
|
} |
77
|
|
|
|
78
|
|
|
/** |
79
|
|
|
* Retrieves all tokens detected in the statement string, in |
80
|
|
|
* the order they were found. |
81
|
|
|
* |
82
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token[] |
83
|
|
|
*/ |
84
|
|
|
public function getTokens() |
85
|
|
|
{ |
86
|
|
|
return $this->tokensOrdered; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
public function hasTokens() : bool |
90
|
|
|
{ |
91
|
|
|
return !empty($this->tokensOrdered); |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* Whether there were any unknown tokens in the statement. |
96
|
|
|
* |
97
|
|
|
* @return bool |
98
|
|
|
*/ |
99
|
|
|
public function hasUnknown() : bool |
100
|
|
|
{ |
101
|
|
|
$unknown = $this->getUnknown(); |
102
|
|
|
|
103
|
|
|
return !empty($unknown); |
104
|
|
|
} |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* Retrieves all unknown content tokens, if any. |
108
|
|
|
* |
109
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token_Unknown[] |
110
|
|
|
*/ |
111
|
|
|
public function getUnknown() |
112
|
|
|
{ |
113
|
|
|
$result = array(); |
114
|
|
|
|
115
|
|
|
foreach($this->tokensOrdered as $token) |
116
|
|
|
{ |
117
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_Unknown) |
118
|
|
|
{ |
119
|
|
|
$result[] = $token; |
120
|
|
|
} |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
return $result; |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
public function getFirstUnknown() : ?Mailcode_Parser_Statement_Tokenizer_Token_Unknown |
127
|
|
|
{ |
128
|
|
|
$unknown = $this->getUnknown(); |
129
|
|
|
|
130
|
|
|
if(!empty($unknown)) |
131
|
|
|
{ |
132
|
|
|
return array_shift($unknown); |
133
|
|
|
} |
134
|
|
|
|
135
|
|
|
return null; |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
public function getNormalized() : string |
139
|
|
|
{ |
140
|
|
|
$parts = array(); |
141
|
|
|
|
142
|
|
|
foreach($this->tokensOrdered as $token) |
143
|
|
|
{ |
144
|
|
|
$string = $token->getNormalized(); |
145
|
|
|
|
146
|
|
|
if($string != '') |
147
|
|
|
{ |
148
|
|
|
$parts[] = $string; |
149
|
|
|
} |
150
|
|
|
} |
151
|
|
|
|
152
|
|
|
return implode(' ', $parts); |
153
|
|
|
} |
154
|
|
|
|
155
|
|
|
/** |
156
|
|
|
* Goes through all tokenization processors, in the order that |
157
|
|
|
* they are defined in the tokenCategories property. This filters |
158
|
|
|
* the statement string, and extracts the tokens contained within. |
159
|
|
|
* |
160
|
|
|
* @param string $statement |
161
|
|
|
* @throws Mailcode_Exception |
162
|
|
|
* |
163
|
|
|
* @see Mailcode_Parser_Statement_Tokenizer_Process |
164
|
|
|
*/ |
165
|
|
|
protected function tokenize(string $statement) : void |
166
|
|
|
{ |
167
|
|
|
$statement = trim($statement); |
168
|
|
|
$tokens = array(); |
169
|
|
|
|
170
|
|
|
foreach($this->tokenCategories as $tokenCategory) |
171
|
|
|
{ |
172
|
|
|
$processor = $this->createProcessor($tokenCategory, $statement, $tokens); |
173
|
|
|
$processor->process(); |
174
|
|
|
|
175
|
|
|
$statement = $processor->getStatement(); |
176
|
|
|
$tokens = $processor->getTokens(); |
177
|
|
|
} |
178
|
|
|
|
179
|
|
|
$this->tokenized = $statement; |
180
|
|
|
$this->tokensOrdered = $tokens; |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
/** |
184
|
|
|
* @param string $id |
185
|
|
|
* @param string $statement |
186
|
|
|
* @param Mailcode_Parser_Statement_Tokenizer_Token[] $tokens |
187
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Process |
188
|
|
|
* @throws Mailcode_Exception |
189
|
|
|
*/ |
190
|
|
|
protected function createProcessor(string $id, string $statement, array $tokens) : Mailcode_Parser_Statement_Tokenizer_Process |
191
|
|
|
{ |
192
|
|
|
$class = 'Mailcode\Mailcode_Parser_Statement_Tokenizer_Process_'.$id; |
193
|
|
|
|
194
|
|
|
$instance = new $class($this, $statement, $tokens); |
195
|
|
|
|
196
|
|
|
if($instance instanceof Mailcode_Parser_Statement_Tokenizer_Process) |
197
|
|
|
{ |
198
|
|
|
return $instance; |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
throw new Mailcode_Exception( |
202
|
|
|
'Unknown statement token.', |
203
|
|
|
sprintf( |
204
|
|
|
'The tokenize class [%s] is not present.', |
205
|
|
|
$class |
206
|
|
|
), |
207
|
|
|
self::ERROR_TOKENIZE_METHOD_MISSING |
208
|
|
|
); |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
/** |
212
|
|
|
* @param string $type |
213
|
|
|
* @param string $matchedText |
214
|
|
|
* @param mixed $subject |
215
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token |
216
|
|
|
*/ |
217
|
|
|
public function createToken(string $type, string $matchedText, $subject=null) : Mailcode_Parser_Statement_Tokenizer_Token |
218
|
|
|
{ |
219
|
|
|
$tokenID = $this->generateID(); |
220
|
|
|
|
221
|
|
|
$class = '\Mailcode\Mailcode_Parser_Statement_Tokenizer_Token_'.$type; |
222
|
|
|
|
223
|
|
|
return new $class($tokenID, $matchedText, $subject, $this->getSourceCommand()); |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
public function appendKeyword(string $name) : Mailcode_Parser_Statement_Tokenizer_Token_Keyword |
227
|
|
|
{ |
228
|
|
|
$name = rtrim($name, ':').':'; |
229
|
|
|
|
230
|
|
|
$token = $this->appendToken('Keyword', $name); |
231
|
|
|
|
232
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_Keyword) |
233
|
|
|
{ |
234
|
|
|
return $token; |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
throw new Mailcode_Exception( |
238
|
|
|
'Invalid token created', |
239
|
|
|
'', |
240
|
|
|
self::ERROR_INVALID_TOKEN_CREATED |
241
|
|
|
); |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
public function removeToken(Mailcode_Parser_Statement_Tokenizer_Token $token) : Mailcode_Parser_Statement_Tokenizer |
245
|
|
|
{ |
246
|
|
|
$keep = array(); |
247
|
|
|
$tokenID = $token->getID(); |
248
|
|
|
|
249
|
|
|
foreach ($this->tokensOrdered as $checkToken) |
250
|
|
|
{ |
251
|
|
|
if($checkToken->getID() !== $tokenID) |
252
|
|
|
{ |
253
|
|
|
$keep[] = $checkToken; |
254
|
|
|
} |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
$this->tokensOrdered = $keep; |
258
|
|
|
|
259
|
|
|
$this->triggerTokensChanged(); |
260
|
|
|
|
261
|
|
|
return $this; |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
/** |
265
|
|
|
* @param string $type |
266
|
|
|
* @param string $matchedText |
267
|
|
|
* @param mixed $subject |
268
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token |
269
|
|
|
*/ |
270
|
|
|
protected function appendToken(string $type, string $matchedText, $subject=null) : Mailcode_Parser_Statement_Tokenizer_Token |
271
|
|
|
{ |
272
|
|
|
$token = $this->createToken($type, $matchedText, $subject); |
273
|
|
|
|
274
|
|
|
$this->tokensOrdered[] = $token; |
275
|
|
|
|
276
|
|
|
$this->triggerTokensChanged(); |
277
|
|
|
|
278
|
|
|
return $token; |
279
|
|
|
} |
280
|
|
|
|
281
|
|
|
/** |
282
|
|
|
* Generates a unique alphabet-based ID without numbers |
283
|
|
|
* to use as token name, to avoid conflicts with the |
284
|
|
|
* numbers detection. |
285
|
|
|
* |
286
|
|
|
* @return string |
287
|
|
|
*/ |
288
|
|
|
protected function generateID() : string |
289
|
|
|
{ |
290
|
|
|
static $alphas; |
291
|
|
|
|
292
|
|
|
if(!isset($alphas)) |
293
|
|
|
{ |
294
|
|
|
$alphas = range('A', 'Z'); |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
$amount = 12; |
298
|
|
|
|
299
|
|
|
$result = ''; |
300
|
|
|
|
301
|
|
|
for($i=0; $i < $amount; $i++) |
302
|
|
|
{ |
303
|
|
|
$result .= $alphas[array_rand($alphas)]; |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
if(!in_array($result, self::$ids)) |
307
|
|
|
{ |
308
|
|
|
self::$ids[] = $result; |
309
|
|
|
return $result; |
310
|
|
|
} |
311
|
|
|
|
312
|
|
|
return $this->generateID(); |
313
|
|
|
} |
314
|
|
|
|
315
|
|
|
/** |
316
|
|
|
* @param callable $callback |
317
|
|
|
*/ |
318
|
|
|
public function onTokensChanged($callback) : void |
319
|
|
|
{ |
320
|
|
|
if(is_callable($callback)) |
321
|
|
|
{ |
322
|
|
|
$this->changeHandlers[] = $callback; |
323
|
|
|
} |
324
|
|
|
} |
325
|
|
|
|
326
|
|
|
protected function triggerTokensChanged() : void |
327
|
|
|
{ |
328
|
|
|
foreach ($this->changeHandlers as $callback) |
329
|
|
|
{ |
330
|
|
|
$callback($this); |
331
|
|
|
} |
332
|
|
|
} |
333
|
|
|
} |
334
|
|
|
|